smart load management of water injection systems in
TRANSCRIPT
Smart load management of water injection systems in offshore oil andgas platforms integrating wind power
Santiago Sanchez1,*, Elisabetta Tedeschi2, Jesus Silva3, M. Jafar4, A. Marichalar5
1Norwegian University of Science and Technology2Norwegian University of Science and Technology3Norwegian University of Science and Technology4DNV GL5DNV GL*[email protected]
Abstract: The high pollution coming from the use of gas turbines in oil and gas (O&G) platforms is
calling for more sustainable solutions. One of those is to use wind turbines (WT) to supply power to
the water injection systems (WIS) of offshore O&G installations and explore the potential of smart
energy management in providing more efficient and green solutions to the O&G sector. The effect
of WT integration into the local power system and the coordination with the local gas turbines,
need to be carefully analysed since the natural intermittency of the wind resource can jeopardize
the stability and efficiency of the offshore grid. In general, the existence of flexible loads interfaced
by variable speed drives (VSD), such as water injection systems, can help overcome some of the
challenges related to wind intermittency and local rotor angle stability: suitable control of such
non-essential loads and load segregation can be implemented to reduce the effect of wind power
fluctuations, balance power generation and consumption and contribute to maintaining the optimal
efficiency of the gas turbine adjusting its loading conditions. The basic assumption is that the
WIS can follow the generation available from the wind turbines. The assumption is reasonable
since WISs do not require a fixed injection rate. This work will investigate the system dynamics
in the event of short-term wind-induced power fluctuations, analysing the evolution of electrical
variables, such as active power and generator rotor speed/frequency, under different operating
conditions, to specifically evaluate the possible arise of low-frequency oscillations. The impact of
an adequate WIS load control to counteract wind variations and increase the system damping is
also further explored. Moreover, in order to prevent the event of critical rotor oscillations in the
gas turbine and other directly-connected rotating machines due to the WT dynamics, an algorithm
for system damping estimation is proposed and its performance studied in the selected test cases.
1. Introduction
The need of sustainability at global level encourages the implementation of greener solutions, espe-
cially in the most polluting industrial sectors. The will of reducing CO2/NOx emissions from oil
and gas installations, along with the difficulty of deploying new gas turbines to supply increasing
loads due to space and structural constraints on the platforms, makes the idea of integrating wind
turbines to O&G attractive. This paper, in particular, investigates the possibility of using wind
power to supply the platform water-injection systems, similarly to the concept proposed by DNV
1
GL in the WinWin project [1]. The objective of the water injection system is to inject raw seawater
or processed water in order to pressurize the well and consequently increase the oil recovery rate
(ORR) using the energy generated locally by the wind turbines. The DNV GL technical assessment
considered a stand-alone WIS, not connected to the platform, where the main subsystem were the
wind turbine and the mechanical set-up composed of a VSD and an induction motor. Here, the
solution proposed is a hybrid configuration, where a floating wind turbine is integrated into the ex-
isting power system based on gas turbines, allowing retrofitting of existing installations. The wind
turbine works in parallel with the gas turbines and it supplies a flexible load that represents a water
injection pump. However, as indicated in [2], there are several challenges associated with wind
power integration that might affect the secure and stable operation of the O&G platform power
system. Those of main interest in this work are the system dynamics due to the wind variability
and power fluctuations affecting the frequency and voltage of the O&G platform.
2. State of the art
The integration of wind power into O&G platforms was investigated only relatively recently, and
still a limited number of contributions is available on the topic. Some of the previous studies [3–5]
targeted electrified O&G installations, in particular those being interconnected to the main power
system through an HVDC link, and excluding the presence of a gas turbines on the platform it-
self. Dynamics and stability issues of such hybrid AC/DC systems, as well as their most relevant
operational scenarios, are inherently different from those of an isolated system, such as the one
analyzed in this paper. Among the studies focusing on an isolated system, [6] performs voltage
and frequency stability analysis checking grid code compliance under two critical scenarios, such
as loss of load and load start-up. Ref. [7] performs similar analyses considering the integration
of a wind farm of up to 100 MW into an O&G cluster: the paper considers additional operational
scenarios, such as loss of generation (by disconnection of the wind farm or of the gas turbine) and
includes loss analyses. Ref. [8] also includes voltage and frequency stability analyses with simi-
lar focus, but extends the range of considered scenarios to load start-up, loss of wind generation
and of interconnection, and includes the quantification of CO2 and NOx emission reduction due
to the wind integration. Unlike the present paper, however, all the above mentioned contributions
do not take into account the effects of the short-term wind power variability. Effect of wind in-
termittency and realistic wind power profiles, have only been considered in [9, 10]. While [9] is
a pure energy analysis, with no transient stability considerations, ref. [10] is the only contribution
targeting voltage and frequency stability analysis under rapidly varying generation condition. As
for [8], such analysis is, however, only aimed at ensuring grid code compliance of the voltage and
frequency profiles, according to the IEC-61892 standard for electrical installations in mobile and
fixed offshore units. To the knowledge of the authors, none of the previous contributions has fo-
cused on another aspect that can be fundamental to ensure the proper operation of wind-powered
O&G installations, i.e. the possible arise of electro-mechanical, low-frequency, oscillations, which
can jeopardize, not only the efficiency, but also the overall stability of the considered system.
The paper is organized as follows: section 3 frames the problem and highlights its relevance.
Section 4 presents the O&G platform power system and its components. Section 5 describes the
scenarios used to test the system, section 6 presents the damping coefficient theory and tests on the
main test cases. Finally, section 7 presents the conclusions of the paper and the appendices present
the parameters of the components used in the O&G power system.
2
3. Relevance of the investigated problem and contributions of the paper
The problem of electro-mechanical oscillations has been widely investigated in traditional large-
scale power systems. Such low frequency oscillations, in a range between 0.1 and 2 Hz, are caused
by instantaneous unbalances between generated and consumed power, and they result in rotor angle
oscillations of the electrical generators connected to the grid [11]. In particular, oscillations in the
range 0.1 to 0.7 Hz are considered the most critical ones, and several cases in which they have led
to protection and generation tripping, and eventually to major blackouts, have been reported [11].
In addition to these system-level consequences of low frequency oscillations, they can provoke ”fa-
tigue of machine shafts, cause excessive wear of mechanical actuators of machine controllers” [12]
and decrease the efficiency of system operation, by affecting most of the interconnected compo-
nents.
In particular, electro-mechanical oscillations can lead to abnormal stresses in, and damage to, the
generator-turbine shaft [13], and, in the considered O&G installation, this could impact the main
gas turbine group.
This phenomenon, however, would not affect only the generators, since it has been shown that
energy oscillations can occur between generators and induction motors [14], and they can be am-
plified under certain conditions, depending on the respective phase of the potential and kinetic
energy of the rotating masses. This can affect all the ac induction motors directly connected to the
O&G distribution system, whose share can vary depending on the specific platform [3].
In [15] it was specifically shown that low-frequency (sub-harmonic) oscillations have an adverse
effect on induction motor performances even if present with very small amplitudes. They poten-
tially lead to reduced speed stability, and eventually result in saturation of the magnetic circuit with
subsequent overcurrents that can damage the machine.
In addition, significant share of the O&G platform power demand is composed of boilers and
heaters [3], which, as thermostatically controlled loads, are considered potential contributors to
low-frequency instability problems [11].
For all the above reasons, and due to the criticality and high reliability of operation required in
O&G platforms, it is considered especially important to assess the possible arise and amplification
of such electro-mechanical oscillations, when intermittent wind power is integrated. The risk of
grid instabilities due of such phenomenon is inherently depended on the frequency of the induced
oscillations, also when their amplitude (and consequently that of related frequency fluctuations) is
very limited [16].
Moreover, while squirrel cage wind generators have a damping effect on the interconnected power
system, this is not true for WT interfaced with a back-to-back converter, as the one considered in
this paper [17]. For wind turbines using full-rated power electronics the effect of wind power on
low frequency oscillations becomes more ambiguous [17] and requires careful evaluation also in
the case of grid interconnected systems.
The scope of the paper is, firstly, to evaluate the impact of the intermittent wind power injection
on the local O&G distribution systems, in terms of induced low-frequency [0.1 -2 Hz] oscillations,
and stress the importance of their early detection. For this purpose, the criticalities of the con-
sidered electric system are first analysed in detail. As mentioned before, main O&G platforms
present standard operation based on synchronous generator (SG) and gas turbine dynamics, as
well as different types of loads, and WT integration can increase the oscillations in the rotor of
3
the synchronous generator and of other directly connected ac motors. The rotor transients with
under-damped oscillations cause the risk that an unstable oscillation appears producing a system
collapse. For this purpose, the damping level of the system is analised both without and with wind
power integration.
Moreover, the possibility of using a smart load management by flexibly controlling the WIS
load to follow the generation is explored as a way to reduce such electro-mechanical oscillations
and evaluate the system damping. WIS motors, if controlled adequately, can play an active role in
counteracting the wind variability and help maintaining the system power balance. As explained
in [18], this type of motors are equipped with variable speed drives, which are flexible units ad-
justable to different operating points or injection rates, and consequently adaptable to multiple
demand levels. The idea proposed here exploits the load flexibility of the WIS to keep a stable and
efficient system operation, ensuring the minimum possible impact to the existing grid, by acting
mainly on controllable and non-critical loads. This concept has been previously explored in [19]
with a focus on voltage and frequency stability, but it is investigated for its possible contribution to
rotor angle stability in the present article.
Once proved the criticality of electro-mechanical oscillations and suggested smart load manage-
ment as mitigation action, this paper proposes, as third contribution, a simplified semi-recursive
algorithm to estimate the damping coefficient of the low-frequency oscillations, which can be used
to set a warning alarm to protect sensitive components in the O&G platform. A discussion on the
challenges of its implementation is also included.
4. Power system for O&G platforms with WIS
The electrical power system of an O&G platform with an integrated wind power generator is
schematically shown in Fig. 1. The system consists of a floating wind turbine, the cable, two
transformers T1 and T2, a gas turbine (GT) with a synchronous generator (SG) at the O&G plat-
form, and subsea WIS based on VSD motor which, in the following analysis, is represented in a
simplified way as a flexible load. A fixed load takes into account the base load consumption of the
platform. The main system components are presented in the following subsections. Additionally,
the proposed strategy is to follow with the flexible load the wind turbine power injection profile;
this can be done by the use of a communications link based on industrial protocols (e.g. the link is
the line between the AC/AC converters and flexible load in Fig. 1).
4.1. Gas turbine (GT)
Gas turbines represent the conventional power supply for O&G platforms, closely integrated with
the electric generator. Their thermal efficiency is low, around 30%. In some cases, they are
equipped with secondary recovery system to extract residual energy normally in the form of heat.
If the wasted heat is recovered, the efficiency can go up to 60% by a steam engine. However, this
is not the common solution found in O&G. In Norway, the greenhouse gas emissions from oil and
gas installations was 15 million tonnes of CO2 equivalent in 2015, corresponding to 28% of the
total domestic emissions [20]. This is driving increasing efforts for their reduction.
The gas turbine and the connected synchronous generator present the configuration shown in Fig.
2. The active power injected into the grid is Pegt, the reactive power is Qegt, the mechanical power
from the GT is Pmgt, the corresponding power reference input of the GT is Pmref , the rotor angular
4
Wind Turbine
AC/AC
ConvertersCable SG
AC pcc
Fixed Load
Flexible Load
GT
T1 T2
Communication link
Power reference to FLEX
Fig. 1. O&G platform with wind turbine and WIS.
speed of the generator is ωr, the field voltage is Efield, the grid voltage at the terminals is Vggt.
The model of the gas turbine is based on the combustion turbine described in [21]: it is a dy-
namic model of first order with time constant Tgt (Fig. 3), so the transfer function of the GT is
Ggt = 1/(Tgts+ 1). The SG used in this paper is the well known simplified synchronous machine
third order model [21–24]. The excitation system is an IEEE type 1 and the model is implemented
according to the recommendations in [23]. The synchronous generator (SG) has transfer func-
tion 1/(2Hs + kdamp), H is the inertia constant in seconds and kdamp is the damping coefficient
of the generator axis [25]. A pure proportional controller (Hcω = HTgt
) is sufficient and it is de-
veloped based on the modulus optimum strategy [26] applied to the open loop transfer function
GgtOL = (1/(Tgts + 1))(1/(2Hs+ kdamp)).
Synchronous
Generator
ωr
GT
PowerController
ExciterController
Pmref
Pe,measured
VggtEfield
Pegt
Qegt
Pmgt
ωmref
Fig. 2. Gas turbine and synchronous generator configuration.
5
Power controller
rotor SG
PmrefPmgt
ωmref +−
ωr
Hcω ++
Pegt,measured
1Tgts+1
+−
Pegt
12Hs+kdamp
ωr
Fig. 3. Gas turbine and synchronous generator mechanical model and control strategy.
4.2. Water injection system
When an oil well is discovered oil flows naturally out of the reservoir during the first exploitation
phase. Primary production usually recovers only 30 to 35 % of the oil in place. As time passes
the pressure starts to decline and the production decreases. The well must then be intervened
for pressure support. This is called secondary recovery or enhanced oil recovery (EOR), and the
objective is to provide energy artificially to boost a declining production in a mature field. There
are two techniques for secondary EOR process: water injection or gas lift. The latter consists
of injecting gas instead of water and it depends on the physical characteristics of the formation.
It is attractive when high pressure gas is available without compression or when gas cost is low.
Compressors are sensitive to downtime and are not suitable for a fluctuating power source [27].
In other cases water injection is preferable. Water injection effectiveness varies according to the
formation characteristics, it can recover anywhere from 5% to 50% of the remaining oil, enhancing
the well economics. When the water content in relation to oil reaches 90 to 99%, the process
becomes uneconomical [27]. Another form of water injection consists of introducing heated water.
This is attractive for reservoirs with heavy oil content, since heat helps decreasing oil viscosity and
making it more fluid [28].
4.2.1. Flexible load (FLEX): In this paper the WIS is represented as a flexible load. A flexible
load is a load whose active and reactive power can change over time according to a predefined
profile. This is modelled by a device that allows independent control of active (PFLEX) and reactive
(QFLEX) power. The corresponding apparent power is SFLEX . In the simulation in this paper the
load is assumed purely active, hence QFLEX is set to zero.
4.3. Fixed load (FL)
The fixed load represents lighting, heating and comfort loads on the platform. It also includes
systems losses and other passive elements. The model representation of a fixed load uses paral-
lel combination of different RLC elements. The resulting impedance is modelled as a constant
impedance load at a given nominal frequency fe. A three-phase balanced load with no capacitive
elements is selected for the considered application [29]. It can be represented as Zload equal to a
resistive load Rload in parallel with an inductive load (jωeLload); where ωe is the electrical speed.
The load consumes active and reactive power; according to the following relationships:
PFL =V 2
Rload
,
QFL =V 2
ωeLload
,
(1)
6
where V is the nominal voltage at the load terminals, PFL and QFL are the active and reactive
power of the fixed passive load, respectively with apparent power SFL.
4.4. Wind power system
The wind turbine considered in this study is a 6 MW wind turbine interfaced to the grid by a back-
to-back power converter, and it resembles the operation of the HyWind floating turbine [30]. The
configuration of the system is shown in Fig. 4, and it is connected to the transformer T1. The wind
speed is represented by vwind, the generator axis rotor speed is ω, the grid converter controller uses
the grid voltage vg, the AC current i and the voltage references e, for power conditioning on the
grid side. The power Pwt is the electric power that the WT injected into the grid. Furthermore,
when the smart load management is applied the Preference is the power reference in pu sent to
the flexible load by a communication link. The corresponding simulation model on the WT side
consists of a current source (representing the wind turbine, the generator, the MPPT and converter
controller). A grid side converter is connected to this system and is used to control the voltage link
DC. The power characteristic of the turbine is based on a cut-in speed of 4 m/s (0.16 pu), nominal
speed of 12 m/s (0.48 pu) and cut-out speed of 25 m/s (1 pu).
Power extraction
Controller (MPPT)
AC
AC
Grid ConverterController
Generator
ω
vwindvg, ie
T1
Preference to FLEX
Pwt injected
Fig. 4. Wind turbine configuration and control.
4.4.1. Rotor side converter control: The main function of the controlled rectifier on the WT
rotor side is to control the rotor speed of the generator for maximum power extraction [31, 32].
For modelling purposes, the WT structure is simplified to a current source supplying a PMSG,
controlled with an MPPT by the converter controller.
In particular, the MPPT is applied for wind speeds between 4 and 12 m/s, while above the
nominal speed the pitch control ensures that the output power is kept constant to the nominal value
up to a wind speed of 25 m/s. Above the cut-out speed the WT is shut down for safety reasons.
As a first step the impact of a variable wind speed profile (vwind in Fig. 5) on the power
extraction (Pwt in Fig. 6) is studied.
7
0 10 20 30 40 50 60
0
0.5
1
time s
v win
dpu
vwind
Fig. 5. Wind speed profile in pu, base wind speed = 25 m/s.
0 10 20 30 40 50 60
0
0.5
1
time s
Pwtpu
Pwt
Fig. 6. Power extracted from the WT in pu, based on the WT rated power (Pbase = 6 MW).
4.4.2. Grid side converter control: The main objective of the grid side inverter is to transfer
maximum power to the local grid by stabilizing the DC link voltage and ensure compliance with
the grid requirements. The grid side converter control diagram is depicted in Fig. 7. The control
system is developed in the dq synchonous reference frame and is standard for this application and
explained in [33]. The voltage in the DC link Vdc is measured, squared and compared with the
corresponding squared reference Vdcref , a PI voltage controller generates a reference current Idrefto be compared with the actual current in the d-axis id as in (2). A similar controller acting on the
q axis current iq is set such that Iqref=0, then only active power is injected and Qref is set to zero.
The corresponding equations are:
ed = Hci(idref − id)− ωeLiq + vgd,
eq = Hci(iqref − iq) + ωeLid + vgq,
idref = Hcv(V2dcref − V 2
dc) + Pout
(2)
Both d and q current references enter a current controller with decoupling terms, in order to
determine the voltage references ed,q, Hci represents the current controller, Hcv is the voltage con-
troller system and Pout is the output power of the DC link. The dq signals are sent to an inverse
Park transformation block to generate the abc frame voltages. The reference angle is extracted
from the grid voltage vg using a PLL block.
8
V 2dcref
+−
v2dc
Hcv ++
−
Pout
id
Hci ++
−
vd
ωeLiq
ed
iqref +−
iq
Hci ++
+
vq
ωeLid
eq
Fig. 7. Grid side converter control strategy.
5. Test cases
After considering a base case in which no wind turbine is integrated, the system will be tested un-
der multiple cases (see Table 1) changing the proportion between fixed and flexible load as shown
in Fig. 8 in order to analyse the impact of the wind power penetration. The base case (Case I)
presents a change in the load ratio SFLEX/SFL in which SFLEX is set to operate at 100% of its
nominal power (3 MW -no flexibility used) and SFL is varied in three steps (η1 in Fig. 8) from 0.3,
0.7 to 0.88 pu (1 pu equal to 25 MVA) and varies from 0 to 22 MVA (Case I). Moreover, the FL
is consuming each of the powers indicated above for the time period τk with k ∈ 1, 2, 3. This
represents a case in which the WIS is running continuously, independently of the local generation
and other (fixed) loads on the platform undergo normal operation variations. Case II describes the
operation as case I but the WT is included. When a smart load management (SLM) is implemented
the WIS is only activated when the WT is producing power i.e. the SFLEX is following the WT
profile (Case III), the loads ratio behaves as ηno−delay in Fig. 8. A final case (Case IV) is analysed
in which the flexible operation of the WIS based on the wind power production is affected by a
delay (1 s) η1s−delay due to the sum of communication and actuators delays.
0 10 20 30 40 50 600
0.2
0.4
0.6
time s
SFLEX/SFL
η1 ηno−delay η1s−delay
Fig. 8. Loads ratio (SFLEX/SFL) for the test cases.
Oscillations of the generator rotor angular speed in the system are studied to obtain an indicator
of the possible destabilizing effects and abnormal stresses due to the WT penetration and FLEX
load dynamics in the grid [34]. The oscillations may pose a serious threat to system security if they
9
Table 1 Cases under test for the O&G platform electrical system.
Case WT GT FL for the
time ranges
(τ1/τ2/τ3)
SLM active
τ1/τ2/τ3FLEX
I (base case) NC 25 MW
installed,
Balancing
load
7.5/17.5/22
MVA
(0.3/0.7/0.88)
pu
no/no/no 3 MW Constant
IIC [0-6]
MW25 MW
installed,
Balancing
load
7.5/17.5/22
MVA
(0.3/0.7/0.88)
pu
no/no/no 3 MW Constant
IIIC [0-6]
MW25 MW
installed,
Balancing
load
7.5/17.5/22
MVA
(0.3/0.7/0.88)
pu
yes/yes/yes [0-3] MW follows wind
IVC [0-6]
MW25 MW
installed,
Balancing
load
7.5/17.5/22
MVA
(0.3/0.7/0.88)
pu
yes/yes/yes [0-3] MW follows wind
with 1 s delay
Note: τ1 is from 10-30 s, τ2 is from 30-45 s and τ3 is from 45-60 s.
NC:Not connected, C:connected
are not controlled properly [35]. As in other works involving multiple generators the oscillations
observed in the system are in a range of [0.1-2] Hz. The frequency spectrum of the oscillations is
presented in the following subsections for each test with the Fourier transform of the rotor angular
speed at the GT-SG, with the goal of highlight the critical low-frequency contributions.
5.1. Case I: constant FLEX (no WT)
The first test is performed on the O&G power system without the integration of the WT. In this test
the voltage in the AC system, the power quantities for GT (i.e. the electrical active Pegt, reactive
Qegt and mechanical power Pmgt of the GT, respectively) are controlled by the GT (Fig. 9). The
power consumption of the WIS is kept constant at its nominal power (3 MW) after being activated
at 10 s. It represents 12% of the nominal GT power. The passive load is switched in three steps
(i.e. three time ranges τ1, τ2 and τ3). In the first part the power consumed by the load is 7.5 MVA
(0.3 pu) with PF=0.9848 (SFL,1). The active power change at the instant SFLEX is activated is
shown in Fig. 9(a), it is important to note that there is no reactive power change and therefore, the
grid voltage has a low disturbance (in Fig. 9(b) at 10 s). A 0.4 pu load with PF=0.8660 is added
at 30 s (SFL,2). The final load added represents 0.18 pu power of the GT with PF=0.9848 and is
activated at 45 s (SFL,3). The main objective is to analyse the behaviour of the system supplied
by the gas turbine only, which is the less environmentally friendly operation, as a base case. The
initial operation point is the system feeding the passive load SFL,1. The activation of the passive
loads causes step changes in the field voltage Efield producing the regulation of the grid voltage
10
10 20 30 40 50 60
0
0.5
1
τ1τ2 τ3
time s
Ppu
Pmgt
Pegt
Qegt
10 20 30 40 50 60
0
0.5
1
τ1 τ2 τ3
time s
Vpu Efield
Vggt
(a) (b)
Fig. 9. System behaviour in the base case (no WT integrated).
a Mechanical power, active and reactive power of the GT.
b Field voltage and grid voltage of the GT.
10 20 30 40 50 600.98
0.99
1
1.01
time s
ωrpu
ωr,pu
0 0.5 1 1.5 20
2
4
6·10−4
Eω = 0.0074
f Hz
ωrpu
ωr,pu
(a) (b)
Fig. 10. Rotor speed of the base case (no WT integrated).
a Time behaviour.
b Frequency components of the oscillation of ωr. The energy of the frequency spectrum for the
rotor signal is Eω.
Vggt (Fig. 9(b) at 30 and 45 s).
The rotor angular speed of the GT generator is presented in Fig. 10(a). It can be seen that
during the transients it drops up to less than 0.99 pu and shows an oscillatory recovery lasting 5-6
seconds. The frequency spectrum of the rotor speed is also presented where the main component
of ωr is ωf=0 (i.e. the nominal rotor angular speed value). The remaining part of the spectrum is
presented in Fig. 10(b), The energy of this spectrum is presented as Eω1 and its meaning will be
addressed in section 5.3 [36].
5.2. Case II: constant FLEX and WT
An overview of the system waveforms under continuous operation of the WIS at maximum power
(fixed SFLEX) is provided in Fig. 11, where the WT power is divided by the system base apparent
power (Sbase: GT-SG generator rated power) and it is represented in pu. The power quantities
1Here, the term energy is used in the sense of signal processing.
11
for GT and WT are represented by Pwt, Pegt, Qegt and Pmgt, respectively. Additionally, the DC
voltage of the grid connected converter of the WT is Vdcwt. The GT balances both the fixed load
and wind variations, and it is the main responsible for the system stability. After 10 s, the WT
and the FLEX load are activated. The gas turbine reduces the generation when the WT produces
nominal power at 20 s. The GT compensates the power variations at 30 s (i.e. activation of SFL,2),
once the WT reduces its injected power. Figure 12 shows the generators rotor angular speed
behaviour in the time domain (Fig. 12(a)) and the respective Fourier spectrum of the rotor speed
(Fig. 12(b)). Moreover, the energy of the spectrum is also presented. Comparing Fig. 12 and
Fig. 10 it can be seen that the intermittent power injected from the WT significantly contributes to
increase the low frequency oscillations in the generator rotor speed (as it is confirmed by comparing
the corresponding spectral energies) and makes it more irregular. It can also be mentioned that
the synchronous GT feeds the reactive load needs as in the first case, since no contribution is
taken from the WT. Even though GTs are flexible units, constant load variations might affect their
performance, efficiency and increase maintenance requirements.
10 20 30 40 50 600
0.5
1
time s
Ppu
Pwt Pmgt
Pegt Qegt
10 20 30 40 50 60
0.6
0.8
1
1.2
time s
Vpu
Efield Vggt Vdcwt
(a) (b)
Fig. 11. System behaviour for Case II (WT integrated, constant FLEX).
a Mechanical power, active and reactive power of the GT and WT. The apparent power base is
Sbase = 25 MVA.
b Field voltage and grid voltage of the GT.
5.3. Case III: FLEX following wind
A new approach is studied in this case: the load is still divided between fixed component FL
(which is activated in steps as in the cases above) and flexible component FLEX (the FLEX load
now changes following the WT power profile). The inductive share of the load is kept constant
for each FL. The flexible load represents an idealization of a water injection pump with smart
load management following the renewable generation. Since the wind profile and the turbine char-
acteristics are the same as in the base case, the wind turbine dynamics are not explained in this
section. The emphasis is given to the gas turbine response under the new condition. One of the
main characteristics is that the starting of the WT and the FLEX smooths the transient at 10 s
(shown in Fig. 13(a)). Furthermore, the smart management of the FLEX load mitigates the effect
of the load operating transients, reducing the low frequency rotor oscillations. This is confirmed
by the corresponding significant reduction in the energy of the spectrum with Eω = 0.0098 versus
Eω = 0.0152 with constant FLEX (see Fig. 14).
12
10 20 30 40 50 60
0.98
0.99
1
1.01
(10.75, 0.9982)
(18.75, 1.003)
time s
ωrpu
10 0.5 1 1.5 2
0
2
4
6·10−4
Eω = 0.0152
f Hz
ωrpu
ωr,pu
(a) (b)
Fig. 12. Rotor speed behaviour for case II (WT integrated, constant FLEX).
a Time behaviour.
b Frequency components of the oscillation of ωr.
10 20 30 40 50 600
0.5
1
time s
Ppu
Pwt Pmgt
Pegt Qegt
10 20 30 40 50 60
0.6
0.8
1
1.2
time s
Vpu
Efield Vggt Vdcwt
(a) (b)
Fig. 13. System behaviour for Case III (WT and FLEX with same profile).
a Mechanical power, electrical power and reactive power of the GT and WT. The apparent power
base is Sbase = 25 MVA.
b Field voltage and grid voltage of the GT.
5.3.1. FLEX following wind profile: effect of communication delays: A drastic delay (1 s)
in the communication and actuation system, resulting in a phase shift between the WT extracted
power and the FLEX load profile, is simulated Fig. 15 and Fig. 16. Industrial communication
protocols and networks are used to control electrical drives which could be affected in this type of
hazardous environments. Additionally, the time response of the actuators or the controllers can be
added to this delay. Hence, the goal of this test is to see any problem in the tracking of the reference
power of the FLEX and its effects in the system. Again the active power sharing is affected by this
disturbance and the rotor experiences variations around the operation point (Fig. 15). Figure 16
shows that the delay produces more oscillations and higher magnitude of the oscillations in the
rotor and directly involving the electrical frequency, showing the criticality of poor WT power
tracking in the following application.
5.4. Angular speed analysis
13
10 20 30 40 50 60
0.98
0.99
1
1.01
(10.3, 1.0006)
(18.5, 1.0014)
time s
ωrpu
ωr,pu
10 0.5 1 1.5 2
0
2
4
6·10−4
Eω = 0.0098
f Hz
ωrpu
ωr,pu
(a) (b)
Fig. 14. Rotor speed behaviour for Case III (WT and FLEX with same profile).
a Time behaviour.
b Frequency components of the oscillation of fe.
Table 2 Energy of the spectrum of the angular speed for all the test cases.
Case Energy ωr
I 0.0074
II 0.0152
III 0.0098
IV 0.0172
The energy of a signal in a finite band is applied on the rotor angular speed frequency spectrum.
This test is carried-out to quantify the degree of influence of the disturbed system: a rotor angular
speed signal with lower energy value shows that the rotor speed is less perturbed (i.e. close to the
nominal value 1 pu). Table 2 shows the energy values for the four cases described above. Case I
represents the simplest behaviour with the lowest energy value. However, the system does not have
the WT integrated. The connection of the WT (Case II) introduces severe rotor speed oscillations
due to intrinsic wind intermittency. It is possible to conclude that by the application of smart load
management the generator rotor angular speed will be less disturbed (e.g. energy value for the
case III is less than the energy for case II). Additionally, a possible problem in the communication
exemplified by a 1 s delay (case IV) presents the larger energy of the signal value.
6. System damping concept
In order to give an index of the criticality of the penetration of the intermittent WT generation and
load dynamics into the O&G isolated electric system, the concept of damping ratio of dynamic
systems is used [35]. It is useful to analyse the response on the power sharing of the system. The
damping estimation quantifies the changes of the damping coefficient in time domain. The infor-
mation conveyed by the damping coefficient can be used to set alarms in critical power systems
such as O&G isolated grid systems. In this case it is studied on the rotor angular speed of the
GT-SG system. The method uses the second order system (3), with y(t) the signal under test (in
14
10 20 30 40 50 600
0.5
1
time s
Ppu
Pwt Pmgt
Pegt Qegt
10 20 30 40 50 60
0.6
0.8
1
1.2
time s
Vpu
Efield Vggt Vdcwt
(a) (b)
Fig. 15. System behaviour for Case IV (WT and FLEX with same profile additionally 1 s delay on
the FLEX reference).
a Mechanical power, active and reactive power of the GT and WT. The apparent power base is
Sbase = 25 MVA.
b Field voltage and grid voltage of the GT.
this case the rotor speed of the generator), u(t) can be a load demand that excites the system.
d2y(t)
dt2+ 2ρωn
dy(t)
dt+ ω2
ny(t) = u(t). (3)
In (3) the damping ratio is ρ, the natural frequency of the system is ωn. Under-damped systems
present a ρ < 1 and they are the characteristic behaviours studied here.
It is well known from dynamic systems theory that the roots of the characteristic equation (3)
are the eigenvalues of the system and they can be obtained as:
λ1,2 = −ρωn ± jωn
√
1− ρ2,
λ1,2 = −α± jωt,(4)
where ωt is the oscillation frequency and is different ωn, ωt = ωn
√
1− ρ2. The real part of λ is
the attenuation, and the damping ratio can be obtained as ρ = −α/√
α2 + ω2t .
6.1. Damping estimation method
A simple estimation method has been employed in this work based on the fact that just one syn-
chronous generator is considered and the rotor speed of this SG is the variable under test. First
a transient event detector algorithm based on the first derivative has been applied [37]. The step
changes in the system are tracked by the first derivative Y1(n) in (5). Y (n) represents the the rotor
speed discrete signal, where n is the discrete sample. Once the absolute value of the derivative
crosses a threshold value then the detector gives a true logic signal which helps to start the storage
of the signal in a buffer with size τ s.
Y1(n) = Y (n)− Y (n− 1) (5)
After all the samples in the window of the oscillation have been stored the mean value is evaluated
and subtracted from the signal in order to obtain a impulse response signal. The least squares
15
10 20 30 40 50 600.98
0.99
1
1.01
time s
ωrpu
ωr,pu
0 0.5 1 1.5 20
2
4
6·10−4
Eω = 0.0172
f Hz
ωrpu
ωr,pu
(a) (b)
Fig. 16. Behaviour of the rotor speed for Case IV (WT and FLEX with same profile additionally 1
s delay on the FLEX reference).
a Time behaviour.
b Frequency components of the oscillation of fe.
algorithm is applied to the samples in order to obtain ρ and ωt as done in [38]. In order to obtain
the fitting of the curve, first, the discrete model of (3) is obtained in (6). In (6) the input has been
set to zero. The parameters of the polynomial have a relation with the continuous parameters of
(3). The least square model to estimate the parameters is y = Hθ, where θ is a vector with the
parameters a1 and a2 to be estimated, H is the matrix with the known terms [36]. In (7) l − 2 is
the length of the sampled data.
yk + a1yk−1 + a2yk−2 = 0 (6)
ykyk−1
...
yk−l
=
yk−1 yk−2
yk−2 yk−3...
...
yk−(l+1) yk−(l+2)
[
a1a2
]
(7)
The damping and oscillation frequency are obtained as the roots of the polynomial formed
with the parameters in (6) and converted to the continuous domain by the use of the Tustin’s
approximation [34]. With the poles in Laplace domain (s+ λ1)(s+ λ2) = 0 the natural frequency
and damping coefficient are calculated based on the definition used in (4).
6.2. Damping estimation for the WIS and WT penetration
The effect of WT and FLEX dynamics is evaluated with the damping ratio estimated from the
GT-SG rotor speed signal. Fig. 17(a) shows the first derivative of the rotor angular speed of the
GT-SG and the threshold used to detect the disturbances. In this case a threshold of 11×10−4 pu
has been used, the window to estimate the damping ratio and the frequency of oscillation is τ = 2.5s, with a sampling time of 0.1 s as used in [34]. Fig. 17(b) presents the steps of the estimated ρand Fig. 17(c) the estimation of the oscillation frequency of the rotor signal (foscillation for case II
and case III), the points represent the estimation detection which triggers the storage of the signal
to estimate. In case III the FLEX is following the profile of the WT generation and it has ideal
16
tracking without delays. Around 18 s a 14% damping is obtained for case III and 7% damping
for case II. After this last window another disturbance is detected around 30 s which is a result
of the SFLEX2 activation (see Fig. 13(a) after 30 s) this results in a ρII = 1.8% for case II and
ρIII = 5.8% for case III, the next power disturbance is detected before 35 s and both damping
values are around 40%. Around 35 s the event is detected and is produced by the maximum power
injection by the WT. It is important to note the reduction of ρII to 17% and ρIII to 22% due to
the nominal power injected by the WT. After 40 s a disturbance is detected and is associated to
the MPPT operation mode of the WT where ρII is having two threshold activations (ρII = 10 and
8.2%), while ρIII = 10% just after 44 s. Last region after 55 s activates the estimation again when
two events are encountered, namely SFLEX3 activation and WT switches from MPPT to nominal
power generation (ρII = 36% and ρIII = 44%). It can be concluded that the use of the SLM shows
better performance in terms of increased damping than the use of a WIS without wind power pro-
file tracking.
Finally, to compare the behavior under a 1 s communication delay between the FLEX and the
WT, the analysis of the corresponding damping coefficients and the oscillations is presented in Fig.
18. It is easy to see the multiple steps on the damping plot Fig. 18(b) which show the work of the
detector. Moreover, the derivative of the rotor speed (Fig. 18(a)) shows the increased disturbance
in the spikes that represent the transients on the rotor speed. The less affected region is between
the 30 s to 50 s, in this part it is shown how ρ is close to 20%. Between 45-55 s there is a large
disturbance that produces a very low ρ which results from the MPPT of the WT. The frequency
estimated for the oscillations of the power are shown in Fig. 18(c), where low frequency distur-
bances (in agreement with [34, 35]) have been estimated.
The latter example further proves the criticality of the delay in the WT power following by the
FLEX load SLM. This increases the interest in additional research on communication-less meth-
ods, which rely only on local measurement at PFLEX , potentially avoiding the problems of the
communication delays. The validation of such techniques is, however, outside the scope of the
paper and is left open for future work.
7. Conclusion
The use of wind power to supply oil and gas installation has been recently regarded as an option
to make the sector less polluting and more sustainable. Previous studies have mostly investigated
the amount of CO2 and NOx emissions that can be avoided by wind integration into O&G plat-
forms or the voltage and frequency stability limit. One aspect that has not been covered so far is
the possible arise of electro-mechanical oscillations in the O&G power system, due to the inher-
ent wind intermittency and consequent higher power imbalances. This paper shows that under the
traditional scenario of inflexible loads, the use of wind power determines a significant increase of
low-frequency oscillations in the range 0.1- 2 Hz, which are considered especially critical in power
systems from the standpoint of rotor angle stability.
On the other hand, the paper considers that the increasing share of controllable loads, (e.g. wa-
ter injections systems) usually connected to variable speed drives in O&G installations, make the
idea of managing the load attractive, in order to keep the system power balance.
Several cases were discussed to test the impact on the power system when a fixed and a combina-
tion of a fixed and flexible load are connected to the platform. The results indicate that controlling
17
10 20 30 40 50 600
1
2
3
4·10−3
time s
|dω dt|pu
| dωII
dt| | dωIII
dt|
(a)
10 20 30 40 50 600
20
40
60
time s
ρ
ρII,% ρIII,%
(b)
10 20 30 40 50 600
1
2
3
time s
f oscilla
tio
nHz
foscillation,II foscillation,III
(c)
Fig. 17. Behaviour of the damping estimation for cases II and III (WT with constant FLex and WT
with FLEX following same profile).
a Derivative of the rotor speed.
b Damping coefficient in percentage.
c Frequency oscillation of the rotor speed transients foscillation in Hz.
the load demand to follow the wind variations, results in an improved response in terms of lower
average rotor speed deviations and better power balance between generation and consumption with
respect to the case of a constant fixed load (See the zoomed area in Fig. 12 and Fig. 14), possi-
bly allowing a larger share of wind penetration. The presented case, with perfect correspondence
between wind generation profile and load control, should be considered as a proof of concept of
the proposed idea. A more realistic situation considering a time delay in the load variations with
18
10 20 30 40 50 600
1
2
3
4·10−3
time s
|dω dt|pu
| dωIV
dt|
(a)
10 20 30 40 50 600
20
40
60
80
100
time s
ρ
ρIV %
(b)
10 20 30 40 50 600
1
2
3
time s
f oscilla
tio
nHz
foscillation,IV
(c)
Fig. 18. Behaviour of the damping estimation forfor delay test (WT and FLEX with same profile
additionally 1 s delay on the FLEX reference).
a Derivative of the rotor speed.
b Damping coefficient in percentage.
c Frequency oscillation of the rotor speed transients foscillation in Hz.
respect to the wind power profile (because communication, actuators and the controllers have a
finite actuation time) has also been presented. The criticality of such delay, which can hinder the
advantages of the flexible load management and worsen the system performance in terms of in-
creased rotor angle oscillations, has been highlighted. On this respect, techniques reducing such
delay, including communication-less methods, where the load control is based on local frequency
measurements rather than on the actual power information received from the WT, can be consid-
19
ered. In addition, this paper has proposed a method that estimates the system damping and rotor
speed oscillations as a consequence of load variations and the effects of the WT power injection.
This recursive algorithm, based on the first derivative of the rotor speed, automatically calculates
the damping levels after each detected event such as load activations. The algorithm is effective
and simpler compared to previous methods using higher order models and it can be used to alert
on critically low damping conditions in sensitive systems such as O&G applications.
8. References
[1] M. Jafar, P. Vaessen, A. Yanushekvich, Y. Fu, R. Marchall, T. Bosman, M. Irvine, and Y. Yang,
“Hybrid grid, towards a hybrid ac dc transmission grid,” DNV GL strategic research and
innovation position paper, no. 2, 2015.
[2] A. R. Ardal, K. Sharifabadi, O. Bergvoll, and V. Berge, “Challenges with integration and
operation of offshore oil amp; gas platforms connected to an offshore wind power plant,” in
2014 Petroleum and Chemical Industry Conference Europe, June 2014, pp. 1–9.
[3] K. M. Lorentzen, A. Atle Rygg, S. Kamran, and T. Marvin, “Integrating offshore wind power
and multiple oil and gas platforms to the onshore power grid using vsc-hvdc technology.” in
Marine Technology Society Journal, vol. 48, 2014, pp. 31–44.
[4] M. L. Kolstad, K. Sharifabadi, A. Rygg Ardal, and T. Undeland, “Grid integration of offshore
wind power and multiple oil and gas platforms.” in OCEANS-Bergen IEEE, 2013.
[5] M. J. I., E. V. Oyslebo, and M. Korpas., “Electrification of offshore petroleum installations
with offshore wind integration.” in Renewable energy 50, 2013, pp. 558–564.
[6] A. A. Rygg, T. Undeland, and K. Sharifabadi, “Voltage and frequency control in offshore
wind turbines connected to isolated oil platform power systems.” in Energy Procedia, vol. 24,
no. 24, 2012, pp. 229–236.
[7] H. G. Svendsen, M. Hadiya, E. Oyslebo, and K. Uhlen, “Integration of offshore wind farm
with multiple oil and gas platforms.” in IEEE PowerTech, 2011, pp. 1–3.
[8] W. He, K. Uhlen, M. Hadiya, Z. Chen, G. Shi, and E. del Rio, “Case study of integrating an
offshore wind farm with offshore oil and gas platforms and with an onshore electrical grid,”
Journal of Renewable Energy, vol. 2013, pp. 1–10, 2013.
[9] W. He, J. Gunnar, A. Tiit, O. Freydar, H. Tor D, M. Korpas, T. Trond, E. Jale, k. Uhlen, and
J. Emil, “The potential of integrating wind power with offshore oil and gas platforms.” in
Wind Engineering 34, 2010, pp. 125–137.
[10] M. Korpas, L. Warland, W. He, and J. O. Tande, “A case-study on offshore wind power supply
to oil and gas rigs.” Energy Procedia, no. 24, pp. 18–26, 2012.
[11] K. Prasertwong, N. Mithulananthan, , and D. Thakur, “Understanding low-frequency oscil-
lation in power systems,” International Journal of Electrical Engineering Education, vol. 3,
no. 47, pp. 248–262, 2010.
[12] J. V. Milanovic, “”damping of the low-frequency oscillations of the generator: dynamic in-
teractions and the effectiveness of the controllers.”,” in IEE Proceedings-Generation, Trans-
mission and Distribution, 2002, pp. 753–760.
20
[13] I. R. Smith, V. V. Vadher, and J. G. Kettleborough, “Small oscillator stability of a synchronous
machine.” in Advances in Power System Control, Operation and Management, APSCOM-91,
International Conference on. IET, 1991.
[14] Y. Yu, Y. Shen, X. Zhang, J. Zhu, and J. Du, “”the load oscillation energy and its effect on low-
frequency oscillation in power system.”,” in Electric Utility Deregulation and Restructuring
and Power Technologies (DRPT), 2015 5th International Conference on. IEEE, 2015.
[15] J. Qazem, Q. Naman, and M. Shamaseen, “Effects of low frequencies on three phase in-
duction motor performance operating in close proximity to rated speed,” Am. J. Applied Sci,
vol. 4, pp. 284–293, 2007.
[16] M. Lasantha and D. Flynn, “Frequency dynamics during high ccgt and wind penetrations,” in
Power Engineering Conference (AUPEC), 2011 21st Australasian Universities. IEEE, 2011.
[17] J. G. Slootweg and W. L. Kling, “The impact of large scale wind power generation on power
system oscillations,” Electric Power Systems Research, pp. 9–20, 2003.
[18] Offshore-Magazine, “2015 worldwide survey of subsea processing: separation, compression
and pumping systems.” in INTECSEA WorleyParsons Group, 2015.
[19] J. Silva, M. Jafar, A. Marichalar, and E. Tedeschi, “Integration of wind power to supply water
injection systems as controllable loads in offshore oil and gas facilities,” in Offshore Energy
& Storage Symposium (OSES 2016), July 2016, pp. 1–9.
[20] S. Norway, “Norway emissions,” http://ssb.no/en/natur-ogmiljo/statistikker/klimagassn/aar-
forelopige/2016-05-20, 2015, [Online; accessed 22-Jun-2016].
[21] N. Anderson, Power system stability. New York: John Wiley & Sons, Inc., 2001.
[22] W. Leonhard, Control of Electrical Drives, 3rd ed. United States of America: Springer,
2001.
[23] J. Machowski, J. W. Bialek, and J. R. Bumby, Power system dynamics, stability and control,
2nd ed. New Delhi: Wiley, 2008.
[24] P. Kundur, power System Stability and Control. United States of America: McGraw-Hill,
1993.
[25] K. Bonfert, Betriebsverhalten der Synchronmaschine. Berlin: Springer-Verlag, 1967.
[26] K.-J. Astrom, PID Controllers: Theory, Design, and Tuning, 1995.
[27] Petrowiki, “Gas lift,” http://petrowiki.org/Gas lift., 2016, [Online; accessed 18-Nov-2016].
[28] Rigzone., “How does water injection work?” http://www.rigzone.com/training/insight.asp?
insight id=341&c id=4, 2015, [Online; accessed 18-Nov-2016].
[29] T. Van-Cutsem and C. Vournas, Voltage stability of electric power systems. Boston: Kluwer
academic publishers, 2001.
[30] Statoil, “Statoil to build the wold’s first floating wind farm: Hywind scotland,” https://https:
//www.statoil.com/en/news/hywindscotland.html, 2017, [Online; accessed 1-Jun-2017].
21
[31] J. A. Baroudi, V. Dinavahi, and A. M. Knight, “A review of power converter topologies for
wind generators,” Renewable Energy, vol. 32, no. 14, pp. 2369 – 2385, 2007.
[32] R. Melicio, V. Mendes, and J. Catalao, “Power converter topologies for wind energy
conversion systems: Integrated modeling, control strategy and performance simulation,”
Renewable Energy, vol. 35, no. 10, pp. 2165 – 2174, 2010. [Online]. Available:
http://www.sciencedirect.com/science/article/pii/S0960148110001114
[33] L. Harnefors, M. Bongiorno, and S. Lundberg, “Input-admittance calculation and shaping for
controlled voltage-source converters,” Industrial Electronics, IEEE Transactions on, vol. 54,
no. 6, pp. 3323 –3334, dec. 2007.
[34] P. Korba, “Real-time monitoring of electromechanical oscillations in power systems: first
findings,” IET Generation, Transmission Distribution, vol. 1, no. 1, pp. 80–88, January 2007.
[35] J. Turunen, J. Thambirajah, M. Larsson, B. C. Pal, N. F. Thornhill, L. C. Haarla, W. W. Hung,
A. M. Carter, and T. Rauhala, “Comparison of three electromechanical oscillation damping
estimation methods,” IEEE Transactions on Power Systems, vol. 26, no. 4, pp. 2398–2407,
Nov 2011.
[36] R. F. Stengel, Optimal control and estimation. New York: Dover, 1994.
[37] N. M. Arzeno, Z. D. Deng, and C. S. Poon, “Analysis of first-derivative based qrs detection
algorithms,” IEEE Transactions on Biomedical Engineering, vol. 55, no. 2, pp. 478–484, Feb
2008.
[38] J. Thambirajah, N. F. Thornhill, and B. C. Pal, “A multivariate approach towards interarea
oscillation damping estimation under ambient conditions via independent component analysis
and random decrement,” IEEE Transactions on Power Systems, vol. 26, no. 1, pp. 315–322,
Feb 2011.
9. Appendices
9.1. Parameters of the WT system
The converter of the wind turbine has power rate of 6 MVA, a rated voltage 3.3 kV a designed
inductor Lwt = 0.1309 pu, with a resistor Rwt = 0.004 pu for a switching frequency of fsw,wt = 4kHz. The DC link capacitor is Cdc,wt = 43.38 µF. The internal PI controller per unit gains are the
proportional gain kpi = 0.556 pu and the integrator gain kii = 5.333 pu. The DC link controller has
a PI structure with per unit proportional gain kp,dc = 1.9052 pu and the integral gain ki,dc = 423.36pu.
9.2. Parameters of the FLEX
The converter of the FLEX has power rated of 3 MVA a rated voltage of 13.472 kV, a designed
inductor Lflex = 0.13 pu, with a resistor Rflex = 0.004 pu for a switching frequency of fsw,wt = 4kHz. The DC link capacitor is Cdc,flex = 0.2636 pu. The internal PI controller per unit gains are
the same as the designed for the WT.
22
9.3. Parameters of the GT and SG
The GT has a first order model with time constant Tgt = 0.25 s, the controller is a PI with just a
proportional gain in pu kp,gt = 14.8 pu. The inertia used in the SG is H = 3.7 s, the per unit values
of the SG are the resistor Rsg = 0.002 pu, the inductive transient reactanceXL,sg = 0.3 pu, the
damping gain is kdamp = 7.0422, pair of poles = 20, electrical frequency fe = 50 Hz, the power
rate is 25 MVA, the line to line base voltage Vbase,L−L = 13.472 kV. The excitation system and
parameters are taken from [23].
9.4. Parameters cable and transformers
The cable resistance is Rc = 0.01273 Ω/km, the reactance is Xc = 0.0734 Ω/km.
The transformer T1 uses the following rates: V1/V2 = (3.3 kV )/(33 kV ), apparent power rate 6
MVA, series resistor R = 0.0031 pu, series reactor XL = 0.1100 pu. The transformer T2 uses
the following rates: V1/V2 = (33 kV )/(11 kV ), apparent power rate 6 MVA, and uses the same
impedance in per unit as T1.
23